You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[bot] Instrument Portkey AI (140,424 weekly downloads) #2482
Portkey AI's official Node.js SDK (portkey-ai) is a drop-in, OpenAI-SDK-shaped client (portkey.chat.completions.create(...)) that routes generation calls through Portkey's gateway to 250+ LLMs, with its own prompt-management and completion-execution surface distinct from the underlying provider SDKs. This execution surface is not instrumented anywhere in this repository.
What's missing
new Portkey({...}) client construction and portkey.chat.completions.create(...) (and streaming variants) — the SDK's primary generation-execution API — has no auto-instrumentation config, wrapper, or e2e scenario in this repo.
None of portkey-ai, Portkey, or PORTKEY_GATEWAY_URL appear anywhere in js/src/, js/src/auto-instrumentations/configs/all.ts, or any e2e/scenarios/* directory. (This is distinct from the already-instrumented openrouter/ai-sdk gateway-style configs, which target different packages.)
unclear — Braintrust publishes a comparison article, "Braintrust vs Portkey" (https://www.braintrust.dev/articles/braintrust-vs-portkey), describing a pattern where Portkey stays in the request path for routing/guardrails while traces are sent to Braintrust "through its SDK or OpenTelemetry integration" — but this describes manual OTel/generic tracing, not a dedicated auto-instrumentation wrapper for the portkey-ai package. Portkey does not appear in Braintrust's main integrations catalog at https://www.braintrust.dev/docs/integrations (full grid enumerated in the companion E2B gap issue).
Activity